Video playback is being increasingly used as a technique for behavioural research. The importance of critically evaluating the effectiveness of video playback is clear, as available video technology is not designed for nonhuman visual systems. We discuss several aspects concerning the perception of video images that could lead to inconclusive or erroneous results. Researchers should verify that behaviour observed in response to video playback is comparable to behaviour observed in response to live animals. We conducted such a verification using live and video playback methods to measure female response to swords of varying lengths in the green swordtail, Xiphophorus helleri. Using both methods, female response appeared to be an increasing function of male sword length. Females did not differ in their response to live and video versions of noncourting, noninteractive males, however, females tended to prefer video playbacks of males with longer swords, a result that has also been found in experiments using live males. These results suggest that females express the same qualitative mating preference, but not necessarily the same quantitative preference, for sword length when viewing video stimuli. Several methodological factors that may contribute to an apparent difference in the strength of the preference are discussed. Despite these differences, both methods produced comparable results; female response to sworded males tended to increase as sword length increased. These experiments demonstrate that video playback is an effective method to measure female preferences accurately in X. helleri and provide an example of how video playback can be evaluated in other species. Copyright 2000 The Association for the Study of Animal Behaviour.